Analytics, Take Two

| 0 Comments | 0 TrackBacks
Every industry has a segment that invariably gets overhyped. In networked security, it was analytics.

Throughout 2007 and 2008, the value proposition on analytics was eclipsed by hype. After lot of early promises fell flat, some vendors at ISC West are attempting to rehabilitate the technology's image by discussing the very real limitations of the technology, and how various approaches can overcome them.

Here, John Whiteman, president, Americas, for ioimage, talks about the importance of knowing what you want in the context of analytics application. Once users can identify the need, sifting through vendors becomes that much easier.



Here, Eric Eaton, chief technology officer of BRS Labs, talks about reason-based analytics as opposed to more traditional rules-based models. The company this week introduced an analytics system for pan-tilt-zoom cameras. The AISight analytics package gradually learns what background and movement is normal in a video frame, stores that information in memory, and applies anomalies against it.



The biggest rap against perimeter detection analytics has been its rate of false alarms, or, as Eaton calls it, "signal-to-noise ratio," borrowing a term from communications engineering. These ratios can be as low as 1 percent, that is, 99 out of 100 analytics alarms are triggered by normal events, said Eaton. He added that getting that ratio to 25 percent would be significant progress, indicating the real world technology still faces challenges.

Late development: AISight won the "Best in Video Analytics" award at the Security Industry Associations's ISC West New Product Showcase.
Every industry has a segment that invariably gets overhyped. In networked security, it was analytics.

Throughout 2007 and 2008, the value proposition on analytics was eclipsed by hype. After lot of early promises fell flat, some vendors at ISC West are attempting to rehabilitate the technology's image by discussing the very real limitations of the technology, and how various approaches can overcome them.

Here, John Whiteman, president, Americas, for ioimage, talks about the importance of knowing what you want in the context of analytics application. Once users can identify the need, sifting through vendors becomes that much easier.



Here, Eric Eaton, chief technology officer of BRS Labs, talks about reason-based analytics as opposed to more traditional rules-based models. The company this week introduced an analytics system for pan-tilt-zoom cameras. The AISight analytics package gradually learns what background and movement is normal in a video frame, stores that information in memory, and applies anomalies against it.



The biggest rap against perimeter detection analytics has been its rate of false alarms, or, as Eaton calls it, "signal-to-noise ratio," borrowing a term from communications engineering. These ratios can be as low as 1 percent, that is, 99 out of 100 analytics alarms are triggered by normal events, said Eaton. He added that getting that ratio to 25 percent would be significant progress, indicating the real world technology still faces challenges.

Late development: AISight won the "Best in Video Analytics" award at the Security Industry Associations's ISC West New Product Showcase.

No TrackBacks

TrackBack URL: http://www.securitysquared.com/cgi-bin/mt/mt-tb.cgi/13

Leave a comment